Note the new Amelia listserv; info at my web page. Comments below; others
may have more...
On Tue, 16 Apr 2002, David Sobek wrote:
Professor King,
I am attempting to use Amelia and have a couple of questions.
1. We are using a pooled-time series data set. Within this data set
we have panels with all values missing. For example, we have no values of
var1 for Nigeria, but we do have values of other variables. Amelia implies
that it cannot impute any values for var1 for Nigeria, is this true?
if Nigeria is an entire variable, then there isn't anything you can do.
If the observations on Nigeria constitute some fraction of observations -
where all the countries are stacked say - then you're ok as long as you
have no dummy variable for Nigeria. if you need the latter, then you have
a problem (lack of data).
2. Even when I deleted the panels where we did
not have any values for
var1, Amelia still gave me a message that it could not impute values
because some cases were missing all values. Do you have any
suggestions as to why I am getting this error?
i'd run the descriptive statistics just to check.
it is also the case, the amelia uses listwise deletion to create starting
values and so if that would leave you with no obs, then you might have to
change the method of computing starting values.
3. Last, I also got an error message that said we
were using too many
variable names and had to set AMvarn = . However, I could not find
that option in the option section.
now that's a good one. maybe my colleagues will have a suggestion. but
probably if you drop some of the variables with so many missing
observations, you'll be ok. I would start with that, in any event. Just
so you can see it work and then perhaps isolate what variable is causing
the problem.
Gary King
: Gary King, King(a)Harvard.Edu
http://GKing.Harvard.Edu :
: Center for Basic Research Direct (617) 495-2027 :
: in the Social Sciences Assistant (617) 495-9271 :
: 34 Kirkland Street, Rm. 2 HU-MIT DC (617) 495-4734 :
: Harvard U, Cambridge, MA 02138 eFax (928) 832-7022 :
Thank you for your time.
Sincerely,
David Sobek
-
amelia mailing list served by Harvard-MIT Data Center
amelia(a)latte.harvard.edu
Subscribe/Unsubscribe:
http://lists.hmdc.harvard.edu/?info=amelia